]> git.lyx.org Git - lyx.git/blobdiff - src/Paragraph.cpp
Natbib authoryear uses (Ref1; Ref2) by default.
[lyx.git] / src / Paragraph.cpp
index b8db16590caaf2eae6882172eeb823ae02168844..72a26006322c93c1066e9b1dacc5462de6c87927 100644 (file)
@@ -43,7 +43,6 @@
 #include "TextClass.h"
 #include "TexRow.h"
 #include "Text.h"
-#include "VSpace.h"
 #include "WordLangTuple.h"
 #include "WordList.h"
 
@@ -70,8 +69,10 @@ using namespace lyx::support;
 namespace lyx {
 
 namespace {
+
 /// Inset identifier (above 0x10ffff, for ucs-4)
 char_type const META_INSET = 0x200001;
+
 }
 
 
@@ -112,7 +113,7 @@ public:
                if (range_.first > pos) {
                        range_.first += offset;
                        range_.last += offset;
-               } else if (range_.last > pos) {
+               } else if (range_.last >= pos) {
                        range_.last += offset;
                }
        }
@@ -130,12 +131,13 @@ private:
 
 class SpellCheckerState {
 public:
-       SpellCheckerState() {
+       SpellCheckerState()
+       {
                needs_refresh_ = true;
                current_change_number_ = 0;
        }
 
-       void setRange(FontSpan const fp, SpellChecker::Result state)
+       void setRange(FontSpan const fp, SpellChecker::Result state)
        {
                Ranges result;
                RangesIterator et = ranges_.end();
@@ -210,20 +212,24 @@ public:
                return empty_;
        }
 
-       bool needsRefresh() const {
+       bool needsRefresh() const
+       {
                return needs_refresh_;
        }
 
-       SpellChecker::ChangeNumber currentChangeNumber() const {
+       SpellChecker::ChangeNumber currentChangeNumber() const
+       {
                return current_change_number_;
        }
 
-       void refreshRange(pos_type & first, pos_type & last) const {
+       void refreshRange(pos_type & first, pos_type & last) const
+       {
                first = refresh_.first;
                last = refresh_.last;
        }
 
-       void needsRefresh(pos_type pos) {
+       void needsRefresh(pos_type pos)
+       {
                if (needs_refresh_ && pos != -1) {
                        if (pos < refresh_.first)
                                refresh_.first = pos;
@@ -238,13 +244,13 @@ public:
                needs_refresh_ = pos != -1;
        }
 
-       void needsCompleteRefresh(SpellChecker::ChangeNumber change_number) {
+       void needsCompleteRefresh(SpellChecker::ChangeNumber change_number)
+       {
                needs_refresh_ = true;
                refresh_.first = 0;
                refresh_.last = -1;
                current_change_number_ = change_number;
        }
-
 private:
        typedef vector<SpellResultRange> Ranges;
        typedef Ranges::const_iterator RangesIterator;
@@ -295,7 +301,7 @@ public:
        /// if the next character is a combining character).
        /// \return whether a surrogate pair was output.
        bool simpleTeXBlanks(OutputParams const &,
-                            otexstream &, TexRow & texrow,
+                            otexstream &,
                             pos_type i,
                             unsigned int & column,
                             Font const & font,
@@ -308,17 +314,17 @@ public:
                           Change const &, Encoding const &, pos_type & i);
 
        /// This could go to ParagraphParameters if we want to.
-       int startTeXParParams(BufferParams const &, otexstream &, TexRow &,
+       int startTeXParParams(BufferParams const &, otexstream &,
                              OutputParams const &) const;
 
        /// This could go to ParagraphParameters if we want to.
-       int endTeXParParams(BufferParams const &, otexstream &, TexRow &,
-                           OutputParams const &) const;
+       bool endTeXParParams(BufferParams const &, otexstream &,
+                            OutputParams const &) const;
 
        ///
        void latexInset(BufferParams const &,
                                   otexstream &,
-                                  TexRow & texrow, OutputParams &,
+                                  OutputParams &,
                                   Font & running_font,
                                   Font & basefont,
                                   Font const & outerfont,
@@ -331,11 +337,13 @@ public:
        ///
        void latexSpecialChar(
                                   otexstream & os,
+                                  BufferParams const & bparams,
                                   OutputParams const & runparams,
                                   Font const & running_font,
                                   Change const & running_change,
                                   Layout const & style,
                                   pos_type & i,
+                                  pos_type end_pos,
                                   unsigned int & column);
 
        ///
@@ -345,6 +353,12 @@ public:
                pos_type i,
                unsigned int & column);
        ///
+       bool latexSpecialT3(
+               char_type const c,
+               otexstream & os,
+               pos_type i,
+               unsigned int & column);
+       ///
        bool latexSpecialTypewriter(
                char_type const c,
                otexstream & os,
@@ -354,6 +368,7 @@ public:
        bool latexSpecialPhrase(
                otexstream & os,
                pos_type & i,
+               pos_type end_pos,
                unsigned int & column,
                OutputParams const & runparams);
 
@@ -376,9 +391,10 @@ public:
        Language * getSpellLanguage(pos_type const from) const;
 
        Language * locateSpellRange(pos_type & from, pos_type & to,
-                                                               SkipPositions & skips) const;
+                                   SkipPositions & skips) const;
 
-       bool hasSpellerChange() const {
+       bool hasSpellerChange() const
+       {
                SpellChecker::ChangeNumber speller_change_number = 0;
                if (theSpellChecker())
                        speller_change_number = theSpellChecker()->changeNumber();
@@ -397,11 +413,16 @@ public:
                speller_state_.setRange(fp, state);
        }
 
-       void requestSpellCheck(pos_type pos) {
-               speller_state_.needsRefresh(pos);
+       void requestSpellCheck(pos_type pos)
+       {
+               if (pos == -1)
+                       speller_state_.needsCompleteRefresh(speller_state_.currentChangeNumber());
+               else
+                       speller_state_.needsRefresh(pos);
        }
 
-       void readySpellCheck() {
+       void readySpellCheck()
+       {
                speller_state_.needsRefresh(-1);
        }
 
@@ -564,16 +585,16 @@ Paragraph::Private::Private(Private const & p, Paragraph * owner,
 
 
 void Paragraph::addChangesToToc(DocIterator const & cdit,
-       Buffer const & buf) const
+       Buffer const & buf, bool output_active) const
 {
-       d->changes_.addToToc(cdit, buf);
+       d->changes_.addToToc(cdit, buf, output_active);
 }
 
 
 bool Paragraph::isDeleted(pos_type start, pos_type end) const
 {
-       LASSERT(start >= 0 && start <= size(), /**/);
-       LASSERT(end > start && end <= size() + 1, /**/);
+       LASSERT(start >= 0 && start <= size(), return false);
+       LASSERT(end > start && end <= size() + 1, return false);
 
        return d->changes_.isDeleted(start, end);
 }
@@ -581,8 +602,8 @@ bool Paragraph::isDeleted(pos_type start, pos_type end) const
 
 bool Paragraph::isChanged(pos_type start, pos_type end) const
 {
-       LASSERT(start >= 0 && start <= size(), /**/);
-       LASSERT(end > start && end <= size() + 1, /**/);
+       LASSERT(start >= 0 && start <= size(), return false);
+       LASSERT(end > start && end <= size() + 1, return false);
 
        return d->changes_.isChanged(start, end);
 }
@@ -627,7 +648,7 @@ void Paragraph::setChange(Change const & change)
 
 void Paragraph::setChange(pos_type pos, Change const & change)
 {
-       LASSERT(pos >= 0 && pos <= size(), /**/);
+       LASSERT(pos >= 0 && pos <= size(), return);
        d->changes_.set(change, pos);
 
        // see comment in setChange(Change const &) above
@@ -639,15 +660,15 @@ void Paragraph::setChange(pos_type pos, Change const & change)
 
 Change const & Paragraph::lookupChange(pos_type pos) const
 {
-       LASSERT(pos >= 0 && pos <= size(), /**/);
+       LBUFERR(pos >= 0 && pos <= size());
        return d->changes_.lookup(pos);
 }
 
 
 void Paragraph::acceptChanges(pos_type start, pos_type end)
 {
-       LASSERT(start >= 0 && start <= size(), /**/);
-       LASSERT(end > start && end <= size() + 1, /**/);
+       LASSERT(start >= 0 && start <= size(), return);
+       LASSERT(end > start && end <= size() + 1, return);
 
        for (pos_type pos = start; pos < end; ++pos) {
                switch (lookupChange(pos).type) {
@@ -681,8 +702,8 @@ void Paragraph::acceptChanges(pos_type start, pos_type end)
 
 void Paragraph::rejectChanges(pos_type start, pos_type end)
 {
-       LASSERT(start >= 0 && start <= size(), /**/);
-       LASSERT(end > start && end <= size() + 1, /**/);
+       LASSERT(start >= 0 && start <= size(), return);
+       LASSERT(end > start && end <= size() + 1, return);
 
        for (pos_type pos = start; pos < end; ++pos) {
                switch (lookupChange(pos).type) {
@@ -717,7 +738,7 @@ void Paragraph::rejectChanges(pos_type start, pos_type end)
 void Paragraph::Private::insertChar(pos_type pos, char_type c,
                Change const & change)
 {
-       LASSERT(pos >= 0 && pos <= int(text_.size()), /**/);
+       LASSERT(pos >= 0 && pos <= int(text_.size()), return);
 
        // track change
        changes_.insert(change, pos);
@@ -746,10 +767,10 @@ void Paragraph::Private::insertChar(pos_type pos, char_type c,
 
 
 bool Paragraph::insertInset(pos_type pos, Inset * inset,
-                                  Change const & change)
+                           Font const & font, Change const & change)
 {
-       LASSERT(inset, /**/);
-       LASSERT(pos >= 0 && pos <= size(), /**/);
+       LASSERT(inset, return false);
+       LASSERT(pos >= 0 && pos <= size(), return false);
 
        // Paragraph::insertInset() can be used in cut/copy/paste operation where
        // d->inset_owner_ is not set yet.
@@ -757,13 +778,15 @@ bool Paragraph::insertInset(pos_type pos, Inset * inset,
                return false;
 
        d->insertChar(pos, META_INSET, change);
-       LASSERT(d->text_[pos] == META_INSET, /**/);
+       LASSERT(d->text_[pos] == META_INSET, return false);
 
        // Add a new entry in the insetlist_.
        d->insetlist_.insert(inset, pos);
 
        // Some insets require run of spell checker
        requestSpellCheck(pos);
+
+       setFont(pos, font);
        return true;
 }
 
@@ -825,8 +848,8 @@ bool Paragraph::eraseChar(pos_type pos, bool trackChanges)
 
 int Paragraph::eraseChars(pos_type start, pos_type end, bool trackChanges)
 {
-       LASSERT(start >= 0 && start <= size(), /**/);
-       LASSERT(end >= start && end <= size() + 1, /**/);
+       LASSERT(start >= 0 && start <= size(), return 0);
+       LASSERT(end >= start && end <= size() + 1, return 0);
 
        pos_type i = start;
        for (pos_type count = end - start; count; --count) {
@@ -848,8 +871,14 @@ int Paragraph::Private::latexSurrogatePair(otexstream & os, char_type c,
        // FIXME: change tracking
        // Is this correct WRT change tracking?
        Encoding const & encoding = *(runparams.encoding);
-       docstring const latex1 = encoding.latexChar(next);
-       docstring const latex2 = encoding.latexChar(c);
+       docstring latex1 = encoding.latexChar(next).first;
+       if (runparams.inIPA) {
+               string const tipashortcut = Encodings::TIPAShortcut(next);
+               if (!tipashortcut.empty()) {
+                       latex1 = from_ascii(tipashortcut);
+               }
+       }
+       docstring const latex2 = encoding.latexChar(c).first;
        if (docstring(1, next) == latex1) {
                // the encoding supports the combination
                os << latex2 << latex1;
@@ -866,7 +895,7 @@ int Paragraph::Private::latexSurrogatePair(otexstream & os, char_type c,
 
 
 bool Paragraph::Private::simpleTeXBlanks(OutputParams const & runparams,
-                                      otexstream & os, TexRow & texrow,
+                                      otexstream & os,
                                       pos_type i,
                                       unsigned int & column,
                                       Font const & font,
@@ -899,8 +928,7 @@ bool Paragraph::Private::simpleTeXBlanks(OutputParams const & runparams,
                     || text_[i - 1] == ':'
                     || text_[i - 1] == '!'))) {
                os << '\n';
-               texrow.newline();
-               texrow.start(owner_->id(), i + 1);
+               os.texrow().start(owner_->id(), i + 1);
                column = 0;
        } else if (style.free_spacing) {
                os << '~';
@@ -966,7 +994,7 @@ int Paragraph::Private::writeScriptChars(otexstream & os,
                // Stop here if there is a font attribute or encoding change.
                if (found && cit != end && prev_font != cit->font())
                        break;
-               docstring const latex = encoding.latexChar(next);
+               docstring const latex = encoding.latexChar(next).first;
                docstring::size_type const b1 =
                                        latex.find_first_of(from_ascii("{"));
                docstring::size_type const b2 =
@@ -1006,7 +1034,6 @@ bool Paragraph::Private::isTextAt(string const & str, pos_type pos) const
 
 void Paragraph::Private::latexInset(BufferParams const & bparams,
                                    otexstream & os,
-                                   TexRow & texrow,
                                    OutputParams & runparams,
                                    Font & running_font,
                                    Font & basefont,
@@ -1018,10 +1045,12 @@ void Paragraph::Private::latexInset(BufferParams const & bparams,
                                    unsigned int & column)
 {
        Inset * inset = owner_->getInset(i);
-       LASSERT(inset, /**/);
+       LBUFERR(inset);
 
        if (style.pass_thru) {
-               inset->plaintext(os.os(), runparams);
+               odocstringstream ods;
+               inset->plaintext(ods, runparams);
+               os << ods.str();
                return;
        }
 
@@ -1049,8 +1078,7 @@ void Paragraph::Private::latexInset(BufferParams const & bparams,
                                os << "\\protect ";
 
                }
-               texrow.newline();
-               texrow.start(owner_->id(), i + 1);
+               os.texrow().start(owner_->id(), i + 1);
                column = 0;
        }
 
@@ -1069,6 +1097,7 @@ void Paragraph::Private::latexInset(BufferParams const & bparams,
        odocstream::pos_type const len = os.os().tellp();
 
        if (inset->forceLTR()
+           && !runparams.use_polyglossia
            && running_font.isRightToLeft()
            // ERT is an exception, it should be output with no
            // decorations at all
@@ -1092,7 +1121,7 @@ void Paragraph::Private::latexInset(BufferParams const & bparams,
        // ArabTeX, though, cannot handle this special behavior, it seems.
        bool arabtex = basefont.language()->lang() == "arabic_arabtex"
                || running_font.language()->lang() == "arabic_arabtex";
-       if (open_font && inset->noFontChange()) {
+       if (open_font && !inset->inheritFont()) {
                bool closeLanguage = arabtex
                        || basefont.isRightToLeft() == running_font.isRightToLeft();
                unsigned int count = running_font.latexWriteEndChanges(os,
@@ -1115,10 +1144,12 @@ void Paragraph::Private::latexInset(BufferParams const & bparams,
                }
        }
 
-       int tmp;
+       int prev_rows = os.texrow().rows();
 
        try {
-               tmp = inset->latex(os, runparams);
+               runparams.lastid = id_;
+               runparams.lastpos = i;
+               inset->latex(os, runparams);
        } catch (EncodingException & e) {
                // add location information and throw again.
                e.par_id = id_;
@@ -1133,9 +1164,8 @@ void Paragraph::Private::latexInset(BufferParams const & bparams,
                                os << '}';
        }
 
-       if (tmp) {
-               texrow.newlines(tmp);
-               texrow.start(owner_->id(), i + 1);
+       if (os.texrow().rows() > prev_rows) {
+               os.texrow().start(owner_->id(), i + 1);
                column = 0;
        } else {
                column += (unsigned int)(os.os().tellp() - len);
@@ -1147,34 +1177,45 @@ void Paragraph::Private::latexInset(BufferParams const & bparams,
 
 
 void Paragraph::Private::latexSpecialChar(otexstream & os,
+                                         BufferParams const & bparams,
                                          OutputParams const & runparams,
                                          Font const & running_font,
                                          Change const & running_change,
                                          Layout const & style,
                                          pos_type & i,
+                                         pos_type end_pos,
                                          unsigned int & column)
 {
-       char_type const c = text_[i];
+       // With polyglossia, brackets and stuff need not be reversed
+       // in RTL scripts (see bug #8251)
+       char_type const c = (runparams.use_polyglossia) ?
+               owner_->getUChar(bparams, i) : text_[i];
 
        if (style.pass_thru || runparams.pass_thru) {
-               if (c != '\0')
-                       // FIXME UNICODE: This can fail if c cannot
-                       // be encoded in the current encoding.
+               if (c != '\0') {
+                       Encoding const * const enc = runparams.encoding;
+                       if (enc && !enc->encodable(c))
+                               throw EncodingException(c);
                        os.put(c);
+               }
                return;
        }
 
+       // TIPA uses its own T3 encoding
+       if (runparams.inIPA && latexSpecialT3(c, os, i, column))
+               return;
        // If T1 font encoding is used, use the special
        // characters it provides.
        // NOTE: some languages reset the font encoding
        // internally
-       if (!running_font.language()->internalFontEncoding()
+       if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
            && lyxrc.fontenc == "T1" && latexSpecialT1(c, os, i, column))
                return;
 
        // \tt font needs special treatment
-       if (running_font.fontInfo().family() == TYPEWRITER_FAMILY
-               && latexSpecialTypewriter(c, os, i, column))
+       if (!runparams.inIPA
+            && running_font.fontInfo().family() == TYPEWRITER_FAMILY
+            && latexSpecialTypewriter(c, os, i, column))
                return;
 
        // Otherwise, we use what LaTeX provides us.
@@ -1241,15 +1282,16 @@ void Paragraph::Private::latexSpecialChar(otexstream & os,
 
        default:
                // LyX, LaTeX etc.
-               if (latexSpecialPhrase(os, i, column, runparams))
+               if (latexSpecialPhrase(os, i, end_pos, column, runparams))
                        return;
 
                if (c == '\0')
                        return;
 
                Encoding const & encoding = *(runparams.encoding);
+               char_type next = '\0';
                if (i + 1 < int(text_.size())) {
-                       char_type next = text_[i + 1];
+                       next = text_[i + 1];
                        if (Encodings::isCombiningChar(next)) {
                                column += latexSurrogatePair(os, c, next, runparams) - 1;
                                ++i;
@@ -1257,20 +1299,50 @@ void Paragraph::Private::latexSpecialChar(otexstream & os,
                        }
                }
                string script;
-               docstring const latex = encoding.latexChar(c);
+               pair<docstring, bool> latex = encoding.latexChar(c);
+               docstring nextlatex;
+               bool nexttipas = false;
+               string nexttipashortcut;
+               if (next != '\0' && next != META_INSET) {
+                       nextlatex = encoding.latexChar(next).first;
+                       if (runparams.inIPA) {
+                               nexttipashortcut = Encodings::TIPAShortcut(next);
+                               nexttipas = !nexttipashortcut.empty();
+                       }
+               }
+               bool tipas = false;
+               if (runparams.inIPA) {
+                       string const tipashortcut = Encodings::TIPAShortcut(c);
+                       if (!tipashortcut.empty()) {
+                               latex.first = from_ascii(tipashortcut);
+                               latex.second = false;
+                               tipas = true;
+                       }
+               }
                if (Encodings::isKnownScriptChar(c, script)
-                   && prefixIs(latex, from_ascii("\\" + script)))
-                       column += writeScriptChars(os, latex,
+                   && prefixIs(latex.first, from_ascii("\\" + script)))
+                       column += writeScriptChars(os, latex.first,
                                        running_change, encoding, i) - 1;
-               else if (latex.length() > 1 && latex[latex.length() - 1] != '}') {
+               else if (latex.second
+                        && ((!prefixIs(nextlatex, '\\')
+                              && !prefixIs(nextlatex, '{')
+                              && !prefixIs(nextlatex, '}'))
+                            || (nexttipas
+                                && !prefixIs(from_ascii(nexttipashortcut), '\\')))
+                        && !tipas) {
                        // Prevent eating of a following
                        // space or command corruption by
                        // following characters
-                       column += latex.length() + 1;
-                       os << latex << "{}";
+                       if (next == ' ' || next == '\0') {
+                               column += latex.first.length() + 1;
+                               os << latex.first << "{}";
+                       } else {
+                               column += latex.first.length();
+                               os << latex.first << " ";
+                       }
                } else {
-                       column += latex.length() - 1;
-                       os << latex;
+                       column += latex.first.length() - 1;
+                       os << latex.first;
                }
                break;
        }
@@ -1305,6 +1377,26 @@ bool Paragraph::Private::latexSpecialT1(char_type const c, otexstream & os,
 }
 
 
+bool Paragraph::Private::latexSpecialT3(char_type const c, otexstream & os,
+       pos_type /*i*/, unsigned int & column)
+{
+       switch (c) {
+       case '*':
+       case '[':
+       case ']':
+       case '\"':
+               os.put(c);
+               return true;
+       case '|':
+               os << "\\textvertline{}";
+               column += 14;
+               return true;
+       default:
+               return false;
+       }
+}
+
+
 bool Paragraph::Private::latexSpecialTypewriter(char_type const c, otexstream & os,
        pos_type i, unsigned int & column)
 {
@@ -1327,7 +1419,10 @@ bool Paragraph::Private::latexSpecialTypewriter(char_type const c, otexstream &
 }
 
 
-bool Paragraph::Private::latexSpecialPhrase(otexstream & os, pos_type & i,
+/// \param end_pos
+///   If [start_pos, end_pos) does not include entirely the special phrase, then
+///   do not apply the macro transformation.
+bool Paragraph::Private::latexSpecialPhrase(otexstream & os, pos_type & i, pos_type end_pos,
        unsigned int & column, OutputParams const & runparams)
 {
        // FIXME: if we have "LaTeX" with a font
@@ -1337,7 +1432,8 @@ bool Paragraph::Private::latexSpecialPhrase(otexstream & os, pos_type & i,
        // "words" for some definition of word
 
        for (size_t pnr = 0; pnr < phrases_nr; ++pnr) {
-               if (!isTextAt(special_phrases[pnr].phrase, i))
+               if (!isTextAt(special_phrases[pnr].phrase, i)
+                   || (end_pos != -1 && i + int(special_phrases[pnr].phrase.size()) > end_pos))
                        continue;
                if (runparams.moving_arg)
                        os << "\\protect";
@@ -1357,30 +1453,37 @@ void Paragraph::Private::validate(LaTeXFeatures & features) const
                Buffer const & buf = inset_owner_->buffer();
                BufferParams const & bp = buf.params();
                Font f;
-               TexRow tr;
+               TexRow texrow;
                // Using a string stream here circumvents the encoding
                // switching machinery of odocstream. Therefore the
                // output is wrong if this paragraph contains content
                // that needs to switch encoding.
                odocstringstream ods;
-               otexstream os(ods);
+               otexstream os(ods, texrow);
                if (is_command) {
-                       ods << '\\' << from_ascii(layout_->latexname());
+                       os << '\\' << from_ascii(layout_->latexname());
                        // we have to provide all the optional arguments here, even though
                        // the last one is the only one we care about.
                        // Separate handling of optional argument inset.
-                       if (layout_->optargs != 0 || layout_->reqargs != 0)
-                               latexArgInsets(*owner_, os, features.runparams(),
-                                       layout_->reqargs, layout_->optargs);
-                       else
-                               os << from_ascii(layout_->latexparam());
+                       if (!layout_->latexargs().empty()) {
+                               OutputParams rp = features.runparams();
+                               rp.local_font = &owner_->getFirstFontSettings(bp);
+                               latexArgInsets(*owner_, os, rp, layout_->latexargs());
+                       }
+                       os << from_ascii(layout_->latexparam());
                }
                docstring::size_type const length = ods.str().length();
                // this will output "{" at the beginning, but not at the end
-               owner_->latex(bp, f, os, tr, features.runparams(), 0, -1, true);
+               owner_->latex(bp, f, os, features.runparams(), 0, -1, true);
                if (ods.str().length() > length) {
-                       if (is_command)
+                       if (is_command) {
                                ods << '}';
+                               if (!layout_->postcommandargs().empty()) {
+                                       OutputParams rp = features.runparams();
+                                       rp.local_font = &owner_->getFirstFontSettings(bp);
+                                       latexArgInsets(*owner_, os, rp, layout_->postcommandargs(), "post:");
+                               }
+                       }
                        string const snippet = to_utf8(ods.str());
                        features.addPreambleSnippet(snippet);
                }
@@ -1388,7 +1491,7 @@ void Paragraph::Private::validate(LaTeXFeatures & features) const
 
        if (features.runparams().flavor == OutputParams::HTML
            && layout_->htmltitle()) {
-               features.setHTMLTitle(owner_->asString(AS_STR_INSETS));
+               features.setHTMLTitle(owner_->asString(AS_STR_INSETS | AS_STR_SKIPDELETE));
        }
 
        // check the params.
@@ -1628,6 +1731,7 @@ void Paragraph::appendChar(char_type c, Font const & font,
        // when appending characters, no need to update tables
        d->text_.push_back(c);
        setFont(d->text_.size() - 1, font);
+       d->requestSpellCheck(d->text_.size() - 1);
 }
 
 
@@ -1648,6 +1752,7 @@ void Paragraph::appendString(docstring const & s, Font const & font,
        for (size_t i = oldsize; i != newsize; ++i) {
                // track change
                d->changes_.insert(change, i);
+               d->requestSpellCheck(i);
        }
        d->fontlist_.set(oldsize, font);
        d->fontlist_.set(newsize - 1, font);
@@ -1680,12 +1785,10 @@ void Paragraph::insertChar(pos_type pos, char_type c,
 
 
 bool Paragraph::insertInset(pos_type pos, Inset * inset,
-                           Font const & font, Change const & change)
+                           Change const & change)
 {
-       bool const success = insertInset(pos, inset, change);
-       // Set the font/language of the inset...
-       setFont(pos, font);
-       return success;
+       Font no_font;
+       return insertInset(pos, inset, no_font, change);
 }
 
 
@@ -1702,7 +1805,7 @@ Font const & Paragraph::getFontSettings(BufferParams const & bparams,
 {
        if (pos > size()) {
                LYXERR0("pos: " << pos << " size: " << size());
-               LASSERT(pos <= size(), /**/);
+               LBUFERR(false);
        }
 
        FontList::const_iterator cit = d->fontlist_.fontIterator(pos);
@@ -1727,9 +1830,9 @@ Font const & Paragraph::getFontSettings(BufferParams const & bparams,
 
 FontSpan Paragraph::fontSpan(pos_type pos) const
 {
-       LASSERT(pos <= size(), /**/);
-       pos_type start = 0;
+       LBUFERR(pos < size());
 
+       pos_type start = 0;
        FontList::const_iterator cit = d->fontlist_.begin();
        FontList::const_iterator end = d->fontlist_.end();
        for (; cit != end; ++cit) {
@@ -1746,8 +1849,8 @@ FontSpan Paragraph::fontSpan(pos_type pos) const
        }
 
        // This should not happen, but if so, we take no chances.
-       // LYXERR0("Paragraph::getEndPosOfFontSpan: This should not happen!");
-       return FontSpan(pos, pos);
+       LYXERR0("Paragraph::fontSpan: position not found in fontinfo table!");
+       LASSERT(false, return FontSpan(pos, pos));
 }
 
 
@@ -1778,7 +1881,7 @@ Font const & Paragraph::getFirstFontSettings(BufferParams const & bparams) const
 Font const Paragraph::getFont(BufferParams const & bparams, pos_type pos,
                                 Font const & outerfont) const
 {
-       LASSERT(pos >= 0, /**/);
+       LBUFERR(pos >= 0);
 
        Font font = getFontSettings(bparams, pos);
 
@@ -1827,16 +1930,23 @@ FontSize Paragraph::highestFontInRange
 char_type Paragraph::getUChar(BufferParams const & bparams, pos_type pos) const
 {
        char_type c = d->text_[pos];
-       if (!lyxrc.rtl_support)
+       if (!lyxrc.rtl_support || !getFontSettings(bparams, pos).isRightToLeft())
                return c;
 
+       // FIXME: The arabic special casing is due to the difference of arabic
+       // round brackets input introduced in r18599. Check if this should be
+       // unified with Hebrew or at least if all bracket types should be
+       // handled the same (file format change in either case).
+       string const & lang = getFontSettings(bparams, pos).language()->lang();
+       bool const arabic = lang == "arabic_arabtex" || lang == "arabic_arabi"
+               || lang == "farsi";
        char_type uc = c;
        switch (c) {
        case '(':
-               uc = ')';
+               uc = arabic ? c : ')';
                break;
        case ')':
-               uc = '(';
+               uc = arabic ? c : '(';
                break;
        case '[':
                uc = ']';
@@ -1857,15 +1967,14 @@ char_type Paragraph::getUChar(BufferParams const & bparams, pos_type pos) const
                uc = '<';
                break;
        }
-       if (uc != c && getFontSettings(bparams, pos).isRightToLeft())
-               return uc;
-       return c;
+
+       return uc;
 }
 
 
 void Paragraph::setFont(pos_type pos, Font const & font)
 {
-       LASSERT(pos <= size(), /**/);
+       LASSERT(pos <= size(), return);
 
        // First, reduce font against layout/label font
        // Update: The setCharFont() routine in text2.cpp already
@@ -1977,8 +2086,7 @@ docstring Paragraph::expandParagraphLabel(Layout const & layout,
        bool const in_appendix = process_appendix && d->params_.appendix();
        docstring fmt = translateIfPossible(layout.labelstring(in_appendix), lang);
 
-       if (fmt.empty() && layout.labeltype == LABEL_COUNTER
-           && !layout.counter.empty())
+       if (fmt.empty() && !layout.counter.empty())
                return tclass.counters().theCounter(layout.counter, lang);
 
        // handle 'inherited level parts' in 'fmt',
@@ -2069,7 +2177,7 @@ bool Paragraph::usePlainLayout() const
 
 bool Paragraph::isPassThru() const
 {
-       return inInset().getLayout().isPassThru() || d->layout_->pass_thru;
+       return inInset().isPassThru() || d->layout_->pass_thru;
 }
 
 namespace {
@@ -2119,13 +2227,12 @@ string const corrected_env(string const & suffix, string const & env,
 }
 
 
-void adjust_row_column(string const & str, TexRow & texrow, int & column)
+void adjust_column(string const & str, int & column)
 {
        if (!contains(str, "\n"))
                column += str.size();
        else {
                string tmp;
-               texrow.newline();
                column = rsplit(str, tmp, '\n').size();
        }
 }
@@ -2134,8 +2241,7 @@ void adjust_row_column(string const & str, TexRow & texrow, int & column)
 
 
 int Paragraph::Private::startTeXParParams(BufferParams const & bparams,
-                                otexstream & os, TexRow & texrow,
-                                OutputParams const & runparams) const
+                       otexstream & os, OutputParams const & runparams) const
 {
        int column = 0;
 
@@ -2184,7 +2290,7 @@ int Paragraph::Private::startTeXParParams(BufferParams const & bparams,
                else
                        output = corrected_env(begin_tag, "flushright", code, lastpar);
                os << from_ascii(output);
-               adjust_row_column(output, texrow, column);
+               adjust_column(output, column);
                break;
        } case LYX_ALIGN_RIGHT: {
                string output;
@@ -2193,13 +2299,13 @@ int Paragraph::Private::startTeXParParams(BufferParams const & bparams,
                else
                        output = corrected_env(begin_tag, "flushleft", code, lastpar);
                os << from_ascii(output);
-               adjust_row_column(output, texrow, column);
+               adjust_column(output, column);
                break;
        } case LYX_ALIGN_CENTER: {
                string output;
                output = corrected_env(begin_tag, "center", code, lastpar);
                os << from_ascii(output);
-               adjust_row_column(output, texrow, column);
+               adjust_column(output, column);
                break;
        }
        }
@@ -2208,16 +2314,13 @@ int Paragraph::Private::startTeXParParams(BufferParams const & bparams,
 }
 
 
-int Paragraph::Private::endTeXParParams(BufferParams const & bparams,
-                              otexstream & os, TexRow & texrow,
-                              OutputParams const & runparams) const
+bool Paragraph::Private::endTeXParParams(BufferParams const & bparams,
+                       otexstream & os, OutputParams const & runparams) const
 {
-       int column = 0;
-
        LyXAlignment const curAlign = params_.align();
 
        if (curAlign == layout_->align)
-               return column;
+               return false;
 
        switch (curAlign) {
        case LYX_ALIGN_NONE:
@@ -2229,13 +2332,12 @@ int Paragraph::Private::endTeXParParams(BufferParams const & bparams,
        case LYX_ALIGN_LEFT:
        case LYX_ALIGN_RIGHT:
        case LYX_ALIGN_CENTER:
-               if (runparams.moving_arg) {
+               if (runparams.moving_arg)
                        os << "\\protect";
-                       column = 8;
-               }
                break;
        }
 
+       string output;
        string const end_tag = "\n\\par\\end";
        InsetCode code = ownerCode();
        bool const lastpar = runparams.isLastPar;
@@ -2248,40 +2350,34 @@ int Paragraph::Private::endTeXParParams(BufferParams const & bparams,
        case LYX_ALIGN_DECIMAL:
                break;
        case LYX_ALIGN_LEFT: {
-               string output;
                if (owner_->getParLanguage(bparams)->babel() != "hebrew")
                        output = corrected_env(end_tag, "flushleft", code, lastpar);
                else
                        output = corrected_env(end_tag, "flushright", code, lastpar);
                os << from_ascii(output);
-               adjust_row_column(output, texrow, column);
                break;
        } case LYX_ALIGN_RIGHT: {
-               string output;
                if (owner_->getParLanguage(bparams)->babel() != "hebrew")
                        output = corrected_env(end_tag, "flushright", code, lastpar);
                else
                        output = corrected_env(end_tag, "flushleft", code, lastpar);
                os << from_ascii(output);
-               adjust_row_column(output, texrow, column);
                break;
        } case LYX_ALIGN_CENTER: {
-               string output;
                output = corrected_env(end_tag, "center", code, lastpar);
                os << from_ascii(output);
-               adjust_row_column(output, texrow, column);
                break;
        }
        }
 
-       return column;
+       return !output.empty() || lastpar;
 }
 
 
 // This one spits out the text of the paragraph
 void Paragraph::latex(BufferParams const & bparams,
        Font const & outerfont,
-       otexstream & os, TexRow & texrow,
+       otexstream & os,
        OutputParams const & runparams,
        int start_pos, int end_pos, bool force) const
 {
@@ -2312,6 +2408,10 @@ void Paragraph::latex(BufferParams const & bparams,
        if (body_pos > 0) {
                // the optional argument is kept in curly brackets in
                // case it contains a ']'
+               // This is not strictly needed, but if this is changed it
+               // would be a file format change, and tex2lyx would need
+               // to be adjusted, since it unconditionally removes the
+               // braces when it parses \item.
                os << "[{";
                column += 2;
                basefont = getLabelFont(bparams, outerfont);
@@ -2328,7 +2428,7 @@ void Paragraph::latex(BufferParams const & bparams,
 
        Encoding const * const prev_encoding = runparams.encoding;
 
-       texrow.start(id(), 0);
+       os.texrow().start(id(), 0);
 
        // if the paragraph is empty, the loop will not be entered at all
        if (empty()) {
@@ -2336,9 +2436,12 @@ void Paragraph::latex(BufferParams const & bparams,
                        os << '{';
                        ++column;
                }
+               if (!style.leftdelim().empty()) {
+                       os << style.leftdelim();
+                       column += style.leftdelim().size();
+               }
                if (allowcust)
-                       column += d->startTeXParParams(bparams, os, texrow,
-                                                   runparams);
+                       column += d->startTeXParParams(bparams, os, runparams);
        }
 
        for (pos_type i = 0; i < size(); ++i) {
@@ -2367,14 +2470,18 @@ void Paragraph::latex(BufferParams const & bparams,
                                ++column;
                        }
 
+                       if (!style.leftdelim().empty()) {
+                               os << style.leftdelim();
+                               column += style.leftdelim().size();
+                       }
+
                        if (allowcust)
                                column += d->startTeXParParams(bparams, os,
-                                                           texrow,
                                                            runparams);
                }
 
-               Change const & change = runparams.inDeletedInset ? runparams.changeOfDeletedInset
-                                                                : lookupChange(i);
+               Change const & change = runparams.inDeletedInset
+                       ? runparams.changeOfDeletedInset : lookupChange(i);
 
                if (bparams.outputChanges && runningChange != change) {
                        if (open_font) {
@@ -2478,8 +2585,7 @@ void Paragraph::latex(BufferParams const & bparams,
                        // latexSpecialChar ignores spaces if
                        // style.pass_thru is false.
                        if (i != body_pos - 1) {
-                               if (d->simpleTeXBlanks(
-                                               runparams, os, texrow,
+                               if (d->simpleTeXBlanks(runparams, os,
                                                i, column, font, style)) {
                                        // A surrogate pair was output. We
                                        // must not call latexSpecialChar
@@ -2501,16 +2607,15 @@ void Paragraph::latex(BufferParams const & bparams,
                // and then split to handle the two modes separately.
                if (c == META_INSET) {
                        if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
-                               d->latexInset(bparams, os,
-                                               texrow, rp, running_font,
+                               d->latexInset(bparams, os, rp, running_font,
                                                basefont, outerfont, open_font,
                                                runningChange, style, i, column);
                        }
                } else {
                        if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
                                try {
-                                       d->latexSpecialChar(os, rp, running_font, runningChange,
-                                               style, i, column);
+                                       d->latexSpecialChar(os, bparams, rp, running_font, runningChange,
+                                                           style, i, end_pos, column);
                                } catch (EncodingException & e) {
                                if (runparams.dryrun) {
                                        os << "<" << _("LyX Warning: ")
@@ -2560,7 +2665,12 @@ void Paragraph::latex(BufferParams const & bparams,
                os << "}]~";
        }
 
-       if (allowcust && d->endTeXParParams(bparams, os, texrow, runparams)
+       if (!style.rightdelim().empty()) {
+               os << style.rightdelim();
+               column += style.rightdelim().size();
+       }
+
+       if (allowcust && d->endTeXParParams(bparams, os, runparams)
            && runparams.encoding != prev_encoding) {
                runparams.encoding = prev_encoding;
                if (!runparams.isFullUnicode())
@@ -2721,6 +2831,22 @@ void Paragraph::simpleDocBookOnePar(Buffer const & buf,
 }
 
 
+namespace {
+void doFontSwitch(vector<html::FontTag> & tagsToOpen,
+                  vector<html::EndFontTag> & tagsToClose,
+                  bool & flag, FontState curstate, html::FontTypes type)
+{
+       if (curstate == FONT_ON) {
+               tagsToOpen.push_back(html::FontTag(type));
+               flag = true;
+       } else if (flag) {
+               tagsToClose.push_back(html::EndFontTag(type));
+               flag = false;
+       }
+}
+}
+
+
 docstring Paragraph::simpleLyXHTMLOnePar(Buffer const & buf,
                                    XHTMLStream & xs,
                                    OutputParams const & runparams,
@@ -2729,64 +2855,317 @@ docstring Paragraph::simpleLyXHTMLOnePar(Buffer const & buf,
 {
        docstring retval;
 
+       // track whether we have opened these tags
        bool emph_flag = false;
        bool bold_flag = false;
-       string closing_tag;
+       bool noun_flag = false;
+       bool ubar_flag = false;
+       bool dbar_flag = false;
+       bool sout_flag = false;
+       bool wave_flag = false;
+       // shape tags
+       bool shap_flag = false;
+       // family tags
+       bool faml_flag = false;
+       // size tags
+       bool size_flag = false;
 
        Layout const & style = *d->layout_;
 
-       if (!runparams.for_toc && runparams.html_make_pars) {
-               // generate a magic label for this paragraph
-               string const attr = "id='" + magicLabel() + "'";
-               xs << html::CompTag("a", attr);
-       }
+       xs.startParagraph(allowEmpty());
 
        FontInfo font_old =
                style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
 
+       FontShape  curr_fs   = INHERIT_SHAPE;
+       FontFamily curr_fam  = INHERIT_FAMILY;
+       FontSize   curr_size = FONT_SIZE_INHERIT;
+       
+       string const default_family = 
+               buf.masterBuffer()->params().fonts_default_family;              
+
+       vector<html::FontTag> tagsToOpen;
+       vector<html::EndFontTag> tagsToClose;
+       
        // parsing main loop
        for (pos_type i = initial; i < size(); ++i) {
                // let's not show deleted material in the output
                if (isDeleted(i))
                        continue;
 
-               Font font = getFont(buf.params(), i, outerfont);
+               Font const font = getFont(buf.masterBuffer()->params(), i, outerfont);
 
                // emphasis
-               if (font_old.emph() != font.fontInfo().emph()) {
-                       if (font.fontInfo().emph() == FONT_ON) {
-                               xs << html::StartTag("em");
-                               emph_flag = true;
-                       } else if (emph_flag && i != initial) {
-                               xs << html::EndTag("em");
-                               emph_flag = false;
+               FontState curstate = font.fontInfo().emph();
+               if (font_old.emph() != curstate)
+                       doFontSwitch(tagsToOpen, tagsToClose, emph_flag, curstate, html::FT_EMPH);
+
+               // noun
+               curstate = font.fontInfo().noun();
+               if (font_old.noun() != curstate)
+                       doFontSwitch(tagsToOpen, tagsToClose, noun_flag, curstate, html::FT_NOUN);
+
+               // underbar
+               curstate = font.fontInfo().underbar();
+               if (font_old.underbar() != curstate)
+                       doFontSwitch(tagsToOpen, tagsToClose, ubar_flag, curstate, html::FT_UBAR);
+       
+               // strikeout
+               curstate = font.fontInfo().strikeout();
+               if (font_old.strikeout() != curstate)
+                       doFontSwitch(tagsToOpen, tagsToClose, sout_flag, curstate, html::FT_SOUT);
+
+               // double underbar
+               curstate = font.fontInfo().uuline();
+               if (font_old.uuline() != curstate)
+                       doFontSwitch(tagsToOpen, tagsToClose, dbar_flag, curstate, html::FT_DBAR);
+
+               // wavy line
+               curstate = font.fontInfo().uwave();
+               if (font_old.uwave() != curstate)
+                       doFontSwitch(tagsToOpen, tagsToClose, wave_flag, curstate, html::FT_WAVE);
+
+               // bold
+               // a little hackish, but allows us to reuse what we have.
+               curstate = (font.fontInfo().series() == BOLD_SERIES ? FONT_ON : FONT_OFF);
+               if (font_old.series() != font.fontInfo().series())
+                       doFontSwitch(tagsToOpen, tagsToClose, bold_flag, curstate, html::FT_BOLD);
+
+               // Font shape
+               curr_fs = font.fontInfo().shape();
+               FontShape old_fs = font_old.shape();
+               if (old_fs != curr_fs) {
+                       if (shap_flag) {
+                               switch (old_fs) {
+                               case ITALIC_SHAPE:
+                                       tagsToClose.push_back(html::EndFontTag(html::FT_ITALIC));
+                                       break;
+                               case SLANTED_SHAPE:
+                                       tagsToClose.push_back(html::EndFontTag(html::FT_SLANTED));
+                                       break;
+                               case SMALLCAPS_SHAPE:
+                                       tagsToClose.push_back(html::EndFontTag(html::FT_SMALLCAPS));
+                                       break;
+                               case UP_SHAPE:
+                               case INHERIT_SHAPE:
+                                       break;
+                               default:
+                                       // the other tags are for internal use
+                                       LATTEST(false);
+                                       break;
+                               }
+                               shap_flag = false;
+                       }
+                       switch (curr_fs) {
+                       case ITALIC_SHAPE:
+                               tagsToOpen.push_back(html::FontTag(html::FT_ITALIC));
+                               shap_flag = true;
+                               break;
+                       case SLANTED_SHAPE:
+                               tagsToOpen.push_back(html::FontTag(html::FT_SLANTED));
+                               shap_flag = true;
+                               break;
+                       case SMALLCAPS_SHAPE:
+                               tagsToOpen.push_back(html::FontTag(html::FT_SMALLCAPS));
+                               shap_flag = true;
+                               break;
+                       case UP_SHAPE:
+                       case INHERIT_SHAPE:
+                               break;
+                       default:
+                               // the other tags are for internal use
+                               LATTEST(false);
+                               break;
                        }
                }
-               // bold
-               if (font_old.series() != font.fontInfo().series()) {
-                       if (font.fontInfo().series() == BOLD_SERIES) {
-                               xs << html::StartTag("strong");
-                               bold_flag = true;
-                       } else if (bold_flag && i != initial) {
-                               xs << html::EndTag("strong");
-                               bold_flag = false;
+
+               // Font family
+               curr_fam = font.fontInfo().family();
+               FontFamily old_fam = font_old.family();
+               if (old_fam != curr_fam) {
+                       if (faml_flag) {
+                               switch (old_fam) {
+                               case ROMAN_FAMILY:
+                                       tagsToClose.push_back(html::EndFontTag(html::FT_ROMAN));
+                                       break;
+                               case SANS_FAMILY:
+                                       tagsToClose.push_back(html::EndFontTag(html::FT_SANS));
+                                       break;
+                               case TYPEWRITER_FAMILY:
+                                       tagsToClose.push_back(html::EndFontTag(html::FT_TYPE));
+                                       break;
+                               case INHERIT_FAMILY:
+                                       break;
+                               default:
+                                       // the other tags are for internal use
+                                       LATTEST(false);
+                                       break;
+                               }
+                               faml_flag = false;
+                       }
+                       switch (curr_fam) {
+                       case ROMAN_FAMILY:
+                               // we will treat a "default" font family as roman, since we have
+                               // no other idea what to do.
+                               if (default_family != "rmdefault" && default_family != "default") {
+                                       tagsToOpen.push_back(html::FontTag(html::FT_ROMAN));
+                                       faml_flag = true;
+                               }
+                               break;
+                       case SANS_FAMILY:
+                               if (default_family != "sfdefault") {
+                                       tagsToOpen.push_back(html::FontTag(html::FT_SANS));
+                                       faml_flag = true;
+                               }
+                               break;
+                       case TYPEWRITER_FAMILY:
+                               if (default_family != "ttdefault") {
+                                       tagsToOpen.push_back(html::FontTag(html::FT_TYPE));
+                                       faml_flag = true;
+                               }
+                               break;
+                       case INHERIT_FAMILY:
+                               break;
+                       default:
+                               // the other tags are for internal use
+                               LATTEST(false);
+                               break;
+                       }
+               }
+
+               // Font size
+               curr_size = font.fontInfo().size();
+               FontSize old_size = font_old.size();
+               if (old_size != curr_size) {
+                       if (size_flag) {
+                               switch (old_size) {
+                               case FONT_SIZE_TINY:
+                                       tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_TINY));
+                                       break;
+                               case FONT_SIZE_SCRIPT:
+                                       tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_SCRIPT));
+                                       break;
+                               case FONT_SIZE_FOOTNOTE:
+                                       tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_FOOTNOTE));
+                                       break;
+                               case FONT_SIZE_SMALL:
+                                       tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_SMALL));
+                                       break;
+                               case FONT_SIZE_LARGE:
+                                       tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGE));
+                                       break;
+                               case FONT_SIZE_LARGER:
+                                       tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGER));
+                                       break;
+                               case FONT_SIZE_LARGEST:
+                                       tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGEST));
+                                       break;
+                               case FONT_SIZE_HUGE:
+                                       tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_HUGE));
+                                       break;
+                               case FONT_SIZE_HUGER:
+                                       tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_HUGER));
+                                       break;
+                               case FONT_SIZE_INCREASE:
+                                       tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_INCREASE));
+                                       break;
+                               case FONT_SIZE_DECREASE:
+                                       tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_DECREASE));
+                                       break;
+                               case FONT_SIZE_INHERIT:
+                               case FONT_SIZE_NORMAL:
+                                       break;
+                               default:
+                                       // the other tags are for internal use
+                                       LATTEST(false);
+                                       break;
+                               }
+                               size_flag = false;
+                       }
+                       switch (curr_size) {
+                       case FONT_SIZE_TINY:
+                               tagsToOpen.push_back(html::FontTag(html::FT_SIZE_TINY));
+                               size_flag = true;
+                               break;
+                       case FONT_SIZE_SCRIPT:
+                               tagsToOpen.push_back(html::FontTag(html::FT_SIZE_SCRIPT));
+                               size_flag = true;
+                               break;
+                       case FONT_SIZE_FOOTNOTE:
+                               tagsToOpen.push_back(html::FontTag(html::FT_SIZE_FOOTNOTE));
+                               size_flag = true;
+                               break;
+                       case FONT_SIZE_SMALL:
+                               tagsToOpen.push_back(html::FontTag(html::FT_SIZE_SMALL));
+                               size_flag = true;
+                               break;
+                       case FONT_SIZE_LARGE:
+                               tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGE));
+                               size_flag = true;
+                               break;
+                       case FONT_SIZE_LARGER:
+                               tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGER));
+                               size_flag = true;
+                               break;
+                       case FONT_SIZE_LARGEST:
+                               tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGEST));
+                               size_flag = true;
+                               break;
+                       case FONT_SIZE_HUGE:
+                               tagsToOpen.push_back(html::FontTag(html::FT_SIZE_HUGE));
+                               size_flag = true;
+                               break;
+                       case FONT_SIZE_HUGER:
+                               tagsToOpen.push_back(html::FontTag(html::FT_SIZE_HUGER));
+                               size_flag = true;
+                               break;
+                       case FONT_SIZE_INCREASE:
+                               tagsToOpen.push_back(html::FontTag(html::FT_SIZE_INCREASE));
+                               size_flag = true;
+                               break;
+                       case FONT_SIZE_DECREASE:
+                               tagsToOpen.push_back(html::FontTag(html::FT_SIZE_DECREASE));
+                               size_flag = true;
+                               break;
+                       case FONT_SIZE_NORMAL:
+                       case FONT_SIZE_INHERIT:
+                               break;
+                       default:
+                               // the other tags are for internal use
+                               LATTEST(false);
+                               break;
                        }
                }
+
                // FIXME XHTML
                // Other such tags? What about the other text ranges?
 
+               vector<html::EndFontTag>::const_iterator cit = tagsToClose.begin();
+               vector<html::EndFontTag>::const_iterator cen = tagsToClose.end();
+               for (; cit != cen; ++cit)
+                       xs << *cit;
+
+               vector<html::FontTag>::const_iterator sit = tagsToOpen.begin();
+               vector<html::FontTag>::const_iterator sen = tagsToOpen.end();
+               for (; sit != sen; ++sit)
+                       xs << *sit;
+
+               tagsToClose.clear();
+               tagsToOpen.clear();
+
                Inset const * inset = getInset(i);
                if (inset) {
                        if (!runparams.for_toc || inset->isInToc()) {
                                OutputParams np = runparams;
+                               np.local_font = &font;
                                if (!inset->getLayout().htmlisblock())
                                        np.html_in_par = true;
                                retval += inset->xhtml(xs, np);
                        }
                } else {
-                       char_type c = d->text_[i];
+                       char_type c = getUChar(buf.masterBuffer()->params(), i);
 
-                       if (style.pass_thru)
+                       if (style.pass_thru || runparams.pass_thru)
                                xs << c;
                        else if (c == '-') {
                                docstring str;
@@ -2814,6 +3193,7 @@ docstring Paragraph::simpleLyXHTMLOnePar(Buffer const & buf,
        }
 
        xs.closeFontTags();
+       xs.endParagraph();
        return retval;
 }
 
@@ -2845,13 +3225,40 @@ bool Paragraph::isLineSeparator(pos_type pos) const
 
 bool Paragraph::isWordSeparator(pos_type pos) const
 {
+       if (pos == size())
+               return true;
        if (Inset const * inset = getInset(pos))
                return !inset->isLetter();
+       // if we have a hard hyphen (no en- or emdash) or apostrophe
+       // we pass this to the spell checker
+       // FIXME: this method is subject to change, visit
+       // https://bugzilla.mozilla.org/show_bug.cgi?id=355178
+       // to get an impression how complex this is.
+       if (isHardHyphenOrApostrophe(pos))
+               return false;
        char_type const c = d->text_[pos];
-       // We want to pass the ' and escape chars to the spellchecker
-       static docstring const quote = from_utf8(lyxrc.spellchecker_esc_chars + '\'');
-       return (!isLetterChar(c) && !isDigitASCII(c) && !contains(quote, c))
-               || pos == size();
+       // We want to pass the escape chars to the spellchecker
+       docstring const escape_chars = from_utf8(lyxrc.spellchecker_esc_chars);
+       return !isLetterChar(c) && !isDigitASCII(c) && !contains(escape_chars, c);
+}
+
+
+bool Paragraph::isHardHyphenOrApostrophe(pos_type pos) const
+{
+       pos_type const psize = size();
+       if (pos >= psize)
+               return false;
+       char_type const c = d->text_[pos];
+       if (c != '-' && c != '\'')
+               return false;
+       int nextpos = pos + 1;
+       int prevpos = pos > 0 ? pos - 1 : 0;
+       if ((nextpos == psize || isSpace(nextpos))
+               && (pos == 0 || isSpace(prevpos)))
+               return false;
+       return c == '\''
+               || ((nextpos == psize || d->text_[nextpos] != '-')
+               && (pos == 0 || d->text_[prevpos] != '-'));
 }
 
 
@@ -2907,9 +3314,9 @@ void Paragraph::changeLanguage(BufferParams const & bparams,
                if (font.language() == from) {
                        font.setLanguage(to);
                        setFont(i, font);
+                       d->requestSpellCheck(i);
                }
        }
-       d->requestSpellCheck(size());
 }
 
 
@@ -2993,7 +3400,8 @@ void Paragraph::forToc(docstring & os, size_t maxlen) const
 }
 
 
-docstring Paragraph::stringify(pos_type beg, pos_type end, int options, OutputParams & runparams) const
+docstring Paragraph::stringify(pos_type beg, pos_type end, int options,
+        OutputParams const & runparams) const
 {
        odocstringstream os;
 
@@ -3002,13 +3410,16 @@ docstring Paragraph::stringify(pos_type beg, pos_type end, int options, OutputPa
                && !d->params_.labelString().empty())
                os << d->params_.labelString() << ' ';
 
+       OutputParams op = runparams;
+       op.for_search = true;
+
        for (pos_type i = beg; i < end; ++i) {
                char_type const c = d->text_[i];
                if (isPrintable(c) || c == '\t'
                    || (c == '\n' && (options & AS_STR_NEWLINES)))
                        os.put(c);
                else if (c == META_INSET && (options & AS_STR_INSETS)) {
-                       getInset(i)->plaintext(os, runparams);
+                       getInset(i)->plaintext(os, op);
                }
        }
 
@@ -3069,8 +3480,7 @@ void Paragraph::setPlainOrDefaultLayout(DocumentClass const & tclass)
 
 Inset const & Paragraph::inInset() const
 {
-       LASSERT(d->inset_owner_, throw ExceptionMessage(BufferException,
-               _("Memory problem"), _("Paragraph not properly initialized")));
+       LBUFERR(d->inset_owner_);
        return *d->inset_owner_;
 }
 
@@ -3143,70 +3553,61 @@ char_type Paragraph::transformChar(char_type c, pos_type pos) const
 }
 
 
-int Paragraph::checkBiblio(Buffer const & buffer)
+bool Paragraph::brokenBiblio() const
+{
+       // there is a problem if there is no bibitem at position 0 or
+       // if there is another bibitem in the paragraph.
+       return d->layout_->labeltype == LABEL_BIBLIO
+               && (d->insetlist_.find(BIBITEM_CODE) != 0
+                   || d->insetlist_.find(BIBITEM_CODE, 1) > 0);
+}
+
+
+int Paragraph::fixBiblio(Buffer const & buffer)
 {
-       // FIXME From JS:
-       // This is getting more and more a mess. ...We really should clean
-       // up this bibitem issue for 1.6.
+       // FIXME: What about the case where paragraph is not BIBLIO
+       // but there is an InsetBibitem?
+       // FIXME: when there was already an inset at 0, the return value is 1,
+       // which does not tell whether another inset has been remove; the
+       // cursor cannot be correctly updated.
 
-       // Add bibitem insets if necessary
        if (d->layout_->labeltype != LABEL_BIBLIO)
                return 0;
 
-       bool hasbibitem = !d->insetlist_.empty()
-               // Insist on it being in pos 0
-               && d->text_[0] == META_INSET
-               && d->insetlist_.begin()->inset->lyxCode() == BIBITEM_CODE;
-
-       bool track_changes = buffer.params().trackChanges;
-
-       docstring oldkey;
-       docstring oldlabel;
-
-       // remove a bibitem in pos != 0
-       // restore it later in pos 0 if necessary
-       // (e.g. if a user inserts contents _before_ the item)
-       // we're assuming there's only one of these, which there
-       // should be.
-       int erasedInsetPosition = -1;
-       InsetList::iterator it = d->insetlist_.begin();
-       InsetList::iterator end = d->insetlist_.end();
-       for (; it != end; ++it)
-               if (it->inset->lyxCode() == BIBITEM_CODE
-                     && it->pos > 0) {
-                       InsetCommand * olditem = it->inset->asInsetCommand();
-                       oldkey = olditem->getParam("key");
-                       oldlabel = olditem->getParam("label");
-                       erasedInsetPosition = it->pos;
-                       eraseChar(erasedInsetPosition, track_changes);
-                       break;
-       }
+       bool const track_changes = buffer.params().trackChanges;
+       int bibitem_pos = d->insetlist_.find(BIBITEM_CODE);
+       bool const hasbibitem0 = bibitem_pos == 0;
 
-       // There was an InsetBibitem at the beginning, and we didn't
-       // have to erase one.
-       if (hasbibitem && erasedInsetPosition < 0)
+       if (hasbibitem0) {
+               bibitem_pos = d->insetlist_.find(BIBITEM_CODE, 1);
+               // There was an InsetBibitem at pos 0, and no other one => OK
+               if (bibitem_pos == -1)
                        return 0;
+               // there is a bibitem at the 0 position, but since
+               // there is a second one, we copy the second on the
+               // first. We're assuming there are at most two of
+               // these, which there should be.
+               // FIXME: why does it make sense to do that rather
+               // than keep the first? (JMarc)
+               Inset * inset = d->insetlist_.release(bibitem_pos);
+               eraseChar(bibitem_pos, track_changes);
+               d->insetlist_.begin()->inset = inset;
+               return -bibitem_pos;
+       }
 
-       // There was an InsetBibitem at the beginning and we did have to
-       // erase one. So we give its properties to the beginning inset.
-       if (hasbibitem) {
-               InsetCommand * inset = d->insetlist_.begin()->inset->asInsetCommand();
-               if (!oldkey.empty())
-                       inset->setParam("key", oldkey);
-               inset->setParam("label", oldlabel);
-               return -erasedInsetPosition;
-       }
-
-       // There was no inset at the beginning, so we need to create one with
-       // the key and label of the one we erased.
-       InsetBibitem * inset =
-               new InsetBibitem(const_cast<Buffer *>(&buffer), InsetCommandParams(BIBITEM_CODE));
-       // restore values of previously deleted item in this par.
-       if (!oldkey.empty())
-               inset->setParam("key", oldkey);
-       inset->setParam("label", oldlabel);
-       insertInset(0, inset,
-                   Change(track_changes ? Change::INSERTED : Change::UNCHANGED));
+       // We need to create an inset at the beginning
+       Inset * inset = 0;
+       if (bibitem_pos > 0) {
+               // there was one somewhere in the paragraph, let's move it
+               inset = d->insetlist_.release(bibitem_pos);
+               eraseChar(bibitem_pos, track_changes);
+       } else
+               // make a fresh one
+               inset = new InsetBibitem(const_cast<Buffer *>(&buffer),
+                                        InsetCommandParams(BIBITEM_CODE));
+
+       insertInset(0, inset, Change(track_changes ? Change::INSERTED 
+                                                  : Change::UNCHANGED));
 
        return 1;
 }
@@ -3343,12 +3744,12 @@ int Paragraph::find(docstring const & str, bool cs, bool mw,
        int i = 0;
        pos_type const parsize = d->text_.size();
        for (i = 0; i < strsize && pos < parsize; ++i, ++pos) {
-               // Ignore ligature break and hyphenation chars while searching
+               // Ignore "invisible" letters such as ligature breaks
+               // and hyphenation chars while searching
                while (pos < parsize - 1 && isInset(pos)) {
-                       const InsetSpecialChar *isc = dynamic_cast<const InsetSpecialChar*>(getInset(pos));
-                       if (isc == 0
-                           || (isc->kind() != InsetSpecialChar::HYPHENATION
-                               && isc->kind() != InsetSpecialChar::LIGATURE_BREAK))
+                       odocstringstream os;
+                       getInset(pos)->toString(os);
+                       if (!getInset(pos)->isLetter() || !os.str().empty())
                                break;
                        pos++;
                }
@@ -3461,16 +3862,13 @@ void Paragraph::locateWord(pos_type & from, pos_type & to,
 
 void Paragraph::collectWords()
 {
-       // This is the value that needs to be exposed in the preferences
-       // to resolve bug #6760.
-       static int minlength = 6;
        pos_type n = size();
        for (pos_type pos = 0; pos < n; ++pos) {
                if (isWordSeparator(pos))
                        continue;
                pos_type from = pos;
                locateWord(from, pos, WHOLE_WORD);
-               if (pos - from >= minlength) {
+               if ((pos - from) >= (int)lyxrc.completion_minlength) {
                        docstring word = asString(from, pos, AS_STR_NONE);
                        FontList::const_iterator cit = d->fontlist_.fontIterator(pos);
                        if (cit == d->fontlist_.end())
@@ -3586,7 +3984,7 @@ Language * Paragraph::Private::getSpellLanguage(pos_type const from) const
 
 void Paragraph::requestSpellCheck(pos_type pos)
 {
-       d->requestSpellCheck(pos == -1 ? size() : pos);
+       d->requestSpellCheck(pos);
 }
 
 
@@ -3638,10 +4036,11 @@ SpellChecker::Result Paragraph::spellCheck(pos_type & from, pos_type & to,
 
        wl = WordLangTuple(word, lang);
 
-       if (!word.size())
+       if (word.empty())
                return result;
 
        if (needsSpellCheck() || check_learned) {
+               pos_type end = to;
                if (!d->ignoreWord(word)) {
                        bool const trailing_dot = to < size() && d->text_[to] == '.';
                        result = speller->check(wl);
@@ -3653,28 +4052,33 @@ SpellChecker::Result Paragraph::spellCheck(pos_type & from, pos_type & to,
                                           word << "\" [" <<
                                           from << ".." << to << "]");
                                } else {
-                                       // spell check with dot appended failed
+                                       // spell check with dot appended failed too
                                        // restore original word/lang value
                                        word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
                                        wl = WordLangTuple(word, lang);
                                }
                        }
                }
-               d->setMisspelled(from, to, result);
+               if (!SpellChecker::misspelled(result)) {
+                       // area up to the begin of the next word is not misspelled
+                       while (end < size() && isWordSeparator(end))
+                               ++end;
+               }
+               d->setMisspelled(from, end, result);
        } else {
                result = d->speller_state_.getState(from);
        }
 
-       bool const misspelled_ = SpellChecker::misspelled(result) ;
-       if (misspelled_ && do_suggestion)
-               speller->suggest(wl, suggestions);
-       else if (misspelled_)
+       if (do_suggestion)
+               suggestions.clear();
+
+       if (SpellChecker::misspelled(result)) {
                LYXERR(Debug::GUI, "misspelled word: \"" <<
                           word << "\" [" <<
                           from << ".." << to << "]");
-       else
-               suggestions.clear();
-
+               if (do_suggestion)
+                       speller->suggest(wl, suggestions);
+       }
        return result;
 }
 
@@ -3729,7 +4133,7 @@ void Paragraph::Private::markMisspelledWords(
 void Paragraph::spellCheck() const
 {
        SpellChecker * speller = theSpellChecker();
-       if (!speller || !size() ||!needsSpellCheck())
+       if (!speller || empty() ||!needsSpellCheck())
                return;
        pos_type start;
        pos_type endpos;
@@ -3763,9 +4167,14 @@ void Paragraph::spellCheck() const
 }
 
 
-bool Paragraph::isMisspelled(pos_type pos) const
+bool Paragraph::isMisspelled(pos_type pos, bool check_boundary) const
 {
-       return SpellChecker::misspelled(d->speller_state_.getState(pos));
+       bool result = SpellChecker::misspelled(d->speller_state_.getState(pos));
+       if (result || pos <= 0 || pos > size())
+               return result;
+       if (check_boundary && (pos == size() || isWordSeparator(pos)))
+               result = SpellChecker::misspelled(d->speller_state_.getState(pos - 1));
+       return result;
 }